These instructions are to determine a 2D ¹H -¹⁵N  shift correlation spectrum (HSQC spectrum)  for a protein sample on the Bruker AMX 500 NMR spectrometer.  Some differences relevant to using the Advance 600 NMR spectrometer are noted.  These instructions assume that you sample has been isotopically labeled by inclusion of ¹⁵N in the growth medium, and that it has not been ¹³C labeled.  If the sample has been ¹³C labeled, see the notes included below about calibration of the ¹³C 90 degree pulse that is needed to effect decoupling of the ¹³Ca  and the ¹³CO from the backbone amide ¹⁵N.  This calibration is required prior to beginning the steps listed below.
 

Version 6/3/02
New reference data sets created 6/02 for the 600.  Check sfo1, sfo3 vs. frequencies posted at the console.

A skeleton of essential steps and especially important points are in bold.
Bulleted items are intended to be explanatory or optional.
Items in brackets are unresolved issues in this draft copy.

Important safety considerations.

• To avoid personal injury from the magnetic field, you should not enter the NMR room if:
• You have a pacemaker or other implanted electromechanical device
• You have done sheet metal work, unless you have confirmed that you are not carrying metal filings in your eyes by an X-ray.
• You have soft tissue metal implants of pins or clamps.
• You have had metal pins implanted in muscle or bone within the last 6 months. Many modern implants are not made of magnetic materials and are safe in the field.  If in doubt, consult a radiologist.
• Leave credit cards, watches, other electronic devices, and all metal objects at the control console table before approaching the magnet.  A metal object that is attracted to and strikes the magnet may quench (damage) it.  If something metal does get stuck to the magnet, it has to be removed without jarring the instrument.  Seek help! If a laptop computer is to be taken into the room, it must not be taken closer to the magnet than the control console.  The 500 MHz magnet is a greater hazard than the 600 MHz magent, because the 600 MHz is self-shielded.
• The magnet is cooled by liquid helium.  Under certain adverse circumstances (quenching - caused by input of kinetic or thermal energy leading to loss of superconduction), there can be a rapid discharge of helium gas from the magnet.  If this happens, leave the room immediately, close the door, and notify the facility manager.  The helium can displace the breathable atmosphere in the room.
• You may not attempt to operate the equipment unless certified by Dr. Hinck to have received the required training.

NMR Sample preparation

The protein sample should have been concentrated to 300 ul, ideally at a concentration of 1 mM.  It should have been added to an 18 mm Shigemi NMR tube with the piston positioned leaving 18 mm of fluid space with no air bubbles.  If the sample is to be loaded at the NMR facility, bring some of the sample buffer in case it needs to be topped off. NB! If the plunger is too low, you are destined to spend several hours trying to shim the instrument, only to have to start over by repositioning the plunger.

Problems set in gradually as the volume drops below 18 mm.  An example is given of a passable shim achieved at 17 mm.  Experiments can be done at 16mm or even lower, but with increasing difficulty in shimming.

General operation of the software

1. The console on the left is for the Avance 500 (replacing the AMX 500); the one on the right for the Avance 600.
2. Ascertain that the instrument is not already in use.
• If another user is conducting an extended experiment, they should have "xlocked" the host Silicon Graphics computer system. The xlock program, which can be started by typing "xlock" from a unix shell, will display a screen saver and prevent access to the system.  Any attempt to exit the screen saver will result in a request for the user's  password.
• If another user is conducting an extended experiment and neglected to "xlock" the system, you can still recognize that the system is in use because:
• They will still be logged in and running xwinnmr on the terminal.
• If the AMX 500 is in use, there will be flashing red and green lights on the NMR RF electronics console in the back-left corner of the room. [Andy: is there such an indicator on the Avance machines???]
• The sample "down" indicator in the middle of the BSMS console (the rectangular panel with lots of machine specific keys sitting to the left of the keyboard for the host computer) will be on indicating that a sample is down in the magnet. (However, some people leave a dummy sample in the magnet between experiments just to keep the dust out).
• The BSMS LOCK indicator will be on.
3. Log in  using your assigned username and password.
• If no one else is logged in, then the only box displayed on the screen will have entry fields for username: and password: .
• The Silicon Graphics computer that you are logging into (designated as the host computer) is a separate computer from the one which directly controls the NMR instrument and gathers the data.  The latter computer, which is hidden inside of the RF console, is called the acquisition computer.  You will interact with the host computer, and it will communicate with the acquisition computer.
• After the experiment described below has been completed, it is possible to first do some preliminary NMR data processing on the host computer using xwinnmr.  Most often, however, once a data set has been collected, it is exported "off line" to one of the other Silicon Graphics computers in the Center's NIS network for processing and analysis.  For  convenience, you will have the same username and password in the NIS system as on the avance500 and avance600.  Accessing data that has been collected on the spectrometer and processing will be described in a separate document.
4. Make an entry in the log book.
5. Open a UNIX shell  by clicking on <desktop><open unix shell> in the small panel of buttons in the upper left corner of the SGI desktop.  Once the shell has opened you will get " 1%" as the unix command line prompt.
6. Type "xwinnmr" to run the software which controls the NMR spectrometer.  The xwinnmr window will open.
 
The following comments explain the general organization of xwinnmr.  Then there will be a section to be completed if this is the first time xwinnmr is being run from this user's directory.  Specific step-by-step instructions for setting up the hsqc experiment start with the "set temperature and load sample" section below.

• All the rest of the commands that you give (except for "xlock") are subcommands of xwinnmr.  There is a command box at the bottom of the xwinnmr window.  When this document says to type "something" to xwinnmr, it means to type "something" in this box followed by the enter key. If the box does not have a blinking cursor, then put the mouse over the xwinnmr window to activate it.
• Some operations will open additional windows.  Most of these can be either closed or left open.  If you close windows, look for a <close>,<quit>, or <return> button in the window.  In some cases, if you close the window with the menu that pops up from the upper left window bar, you will leave the process running, but without an interface.  If you think you have done this, you can use the upper xwinnmr menu <display> <active commands> <show and allow for killing> to see if processes are running without windows, and to kill them (by clicking on them in this menu).
• To govern which windows are on top when they overlap, place the cursor on the exposed edge of a window and right click.  Use "raise" or "lower" on the popup menu to bring the window to the front or put it behind other windows.
• For some windows that use the vi editor to edit files will close when you give the vi command to quit (:q! or :wq).  Remember, if you want to save your edits, to write before quiting (:w, :w <filename>, or :wq). The xwinnmr window should remain open for the duration of your experiment. The unix shell window from which xwinnmr was opened should also remain opened.  You can minimize open windows to icons with the small dot that is 2nd to right on the top window bar, or by choosing <minimize> from the popup menu from the bar in the upper left corner of the window.
• This document will use the syntax <something> to mean to left click on a menu, menu item, or button by the name "something" in the xwinnmr window, or one of the windows opened by xwinnmr commands.  Instructions may specifiy which window the <something> button is in.  Else the <something> is in the active window (the one just opened, or the one last accessed.  Usually putting the mouse over an inactive window is sufficient to activate it.
• To set up your experiment, you will use xwinnmr to manipulate things of the following description: pulse programs, parameters, and datasets.  The overall organization of these things is: that a dataset contains files full of parameters.  One of the parameters points to a file containing a pulse program.  The pulse program accesses the other parameters and runs the acquisition of data.  As a result, data is recorded within the dataset.  You will have to understand something of this organization, because you will need to customize various aspects of the dataset, parameters, and pulse program in order to start your experiment.
• The basic purpose of xwinnmr is to run a "pulse program".  A pulse program is a timed sequence of commands that tells the spectrometer hardware when and how to apply a series of RF pulses, when and how to apply pulsed field grandients, and how the data should be collected. A pulse program may have up to 31 different kinds of pulses and will loop repeatedly through a sequence of pulses.  Each passage through the loop is called a "scan".
• An elementary pulse command would specify the duration of the pulse, and the power to be applied.  For example "p1" is a command within a pulse program to issue a particular kind of pulse named p1.  The duration of pulse p1 is an adjustable parameter also named "p1" and the associated power level is a parameter named hl1.  In the setup procedure for this hsqc pulse program described below, you will have to make some preliminary measurements to determine the proper value for p1 and similar parameters and then supply that information to xwinnmr.
• Commands to set parameters -- xwinnmr provides special commands for setting p1 and other parameters.  The command to enter p1 is also "p1".  In general, you can set a parameter by typing its name as a command.  Alternatively, there is a command ("ased") to bring up the full list of parameters used by the pulse program which can be edited in place.  Whenever one types a command such as "p1" without a value, the program will suggest the previous value used.  One can accept this value by typing enter, or type in a new value followed by enter.  Alternatively, one can just type "p1 newvalue" followed by enter.
• The parameters associated with an experiment are stored in parameter files within the dataset for the experiment.  The action of setting up your dataset copies the parameter files from a prior experiment.  This will cause some of the parameters to be set properly and you won't have to change them.  Others will be set only approximately, and you will have to adjust them after conducting some calibration measurements.  The parameter files are automatically updated each time you adjust a parameter.  You do not have to explicitly save your changes to the parameters.
• You will have to set a few parameters for the pulse program that "ased" fails to list. There are other parameters in the parameter file that are not used by the pulse program (eg. processing parameters, and the temperature parameter for the temperature controller).  You can view all parameters with the command "eda".
• Datasets: When you create a dataset, you specify a name and experiment number.  The dataset name is actually the name of a subdirectory that will be created in your user directory.  When you first create it, it will have nothing but some parameter files (acqu, acqus, acqu2, acqu2s) and a subdirectory named /pdata for actual data storage. The experiment number is used as a name for a subdirectory under /pdata.  You can put more than one experiment in the same dataset, as long as you give them different experiment numbers.  Alternatively, you can keep experiments separate by giving them different dataset names. Generally, experiments that are analyzed as a series (such as a relaxation curve) are placed in a single dataset.  Otherwise, put each new experiment in a new dataset. Typically, people have used the same name as the pulse program for the dataset, although it may be better practice to incorporate both the name of the pulse program and the name of your protein into the dataset name.
• The values of the parameters are kept separate from the pulse program itself. In principle, this allows you to modify the function of the pulse program without having to edit the program itself or without having to know the detailed syntax of the pulse programming language.  However, there are things that have to be set up in the pulse programs that aren't determined by parameters, so you will also have to edit the pulse program file itself.
• In order to edit a pulse program file, you will have to have ownership of the file.  The first time you use a pulse program, you will have to make a copy for which you have ownership.  In subsequent experiments, you can re-edit your own copy of the file. Specific information on how to load, edit, and save pulse program files is given where they appear below.  Alternatively, it is possible to use unix commands within a unix shell to copy and edit pulse programs.  However, be warned that xwinnmr locates these files in system directories, not in your user directory.  It is very easy to edit the wrong files.  The command to specify the default unix editor is <desktop><customize><desktop><utilities>.
• There are other files used by the pulse program and referenced as parameters.  In the calibration procedure below, you will have to enter the names of some of these, but not edit them.  In more general circumstances these might be edited, in which case the same ownership issue would arise.  If you attempt to find these with unix commands, you will need to consult the "files" section of the online manual to find out in which xwinnmr system directory each is located.

If xwinnmr has never been run from this user directory before:

If xwinnmr has never been run from this user directory  before, then there are a few things that will need to be set up one time only:
1. Tell xwinnmr about the available disk units.
1. On upper menu bar <file><search> <edit><Edit Dir List>
2. If /u /hinck and /nall are not all present on the list of disk units, then add them using the commands provided.
3. <Save as default>
2. Set up the user interface.
1. On upper menu bar <display> <options> <userinterface>
2. Set the editor to vi.  You can use one of the other editors if you are already comfortable with them, but the editing commands won't correspond to this document.
3. Set "userinterface" to "extended".  This will cause xwinnmr to give you a little more information as you go.
4. Set "ZGsafety" to off.  When on, this feature asks for confirmation when you are about to overwrite a dataset.   The calibration procedure will generate dozens of these messages if you leave this on.
1. Exit the window.

Set temperature and load sample

The sample is loaded first because it is required to make a number of fine adjustments to the instrument that vary with each individual sample.  The sample will come to thermal equilibrium during these adjustments.  The final shimming should be done after the sample has reached thermal equilibrium, which may require 30 min. if the sample is being warmed from an ice bath.
 
1. With the cursor in the newly opened xwinnmr window, type "edte" to open the edit temperature dialogue box. Click <change> and type in your desired temperature in degrees Kelvin (centigrade +273). This operation sends the temperature directly to the temperature controller and bypasses the temperature parameter in the database.
• The temperature reported by the temperature controller differs from the true temperature by a small amount that may drift with time.  See Temperature calibration and chemical shift referencing to see how to achieve an accurate and constant temperature across all of your experiments.
• Alternatively, you can use a standard parameter setting command,  "te", to set the temperature parameter in the database. To actually send the parameter to the temperature controller, you must then type another command, "teset".   This method requires that you have loaded a database for which you have ownership.  If you get a write permission error, then use the other method.
• Note that you will subsequently load datasets that may change the temperature parameter.  However, this will have no effect unless a command is issued to send the temperature parameter to the temperature controller.
2. Put the sample tube in the float-like holder and adjust the tube to center it in plastic model of the RF coil  supplied.
• There is a plastic holder and a ceramic holder.  Use the ceramic holder if you will work at temperatures greater than 37 C.
3. Depress the "Lift On Off" button on the upper left corner of the BSMS to make the light on the key come on.
• This is a toggle key; pressing it alternates the lift between on and off.
• When the light is on, there is an air current sent through the bore of the machine that will lift a sample float to the top of the machine if one is present.
4. Suspend your sample float in the air current coming from the bore of the spectrometer at the top of the machine.
• If the air current will not hold the sample float at the top of the machine, then something is wrong.  Do not just drop the sample into the bore of the machine.
5. Toggle the "LIFT ON OFF" key off to lower the sample into the magnet.
• Sometimes it is necessary to gently tap the top of the sample tube as it begins to lower to prevent it from sticking.
• If the machine is unsuccessful at lowering the sample, it will automatically cycle the air on and off again in an attempt to lower the sample. It is usually successful on the 2nd try.
• If the BSMS issues the message "SLBC lid close. press standby", the machine thinks that there is a lid over the bore hole.  You probably have tape or markings on your tube just over the float that are being detected by the lid sensor.  Press STANDBY; remove your sample; get the lift back on; try again.  If you tap the tube right after turning the lift off, it should lower before the error is sensed.
6. Check control panel for "down" light on (in the middle of the panel; not on the lift key) to confirm that sample has been lowered into position.  If it doesn't come on, then lift the sample back to the top, clean the tube and holder with ethanol, reposition it, and try lowering it again.

Set the deuterium lock signal:

A separate transmitter and receiver within the instrument will track the deuterium signal from the 5% D₂O in your sample buffer.  This system will detect instrument drift throughout the acquisition and make compensating adjustments.

Make sure that the LOCK key on the BSMS is off and the SWEEP key is on.

On the xwinnmr menu bar click <Windows> and <Lockdisp> to open the lock display window.

Type "lock d2o" to XwinNMR.


This starts a series of automated optimizations of the deuterium detection.  The system then automatically engages the lock as indicated by the LOCK light on the BSMS becoming illuminated.  The lock display window shows the status of the lock signal.  This varies from a flat line immediately after locking to an oscillating line during signal acquisition.  Some operators leave this display on the screen to assure them of a continuing strong lock signal.  It can also be reopend from the menu bar at any time.

Manually setting the deuterium lock:

This operation is not expected to be needed to be needed since the introduction of the lock d2o function.  However, if there appears to be a failure of the automated locking function, running through the manual setting might reveal the problem.

The first section describes the general operation of the BSMS unit, which is also used in manual shimming and for pulse calibration.

General BSMS operation.  This is one of several operations using the BSMS controls.  The general use of the BSMS is described first, followed by the specific instructions for setting the deuterium lock.
 

• The keys on the BSMS are toggle switches.  Pressing a key turns a function on, and then pressing it again turns it back off. There is a light on the key to indicate whether the function is on or off.  A BSMS key will be indicated in caps in this document.
• Toggling on some of the keys allows you to adjust a value. Read carefully; toggling the key back off cancels the adjustment! The value is adjusted by rotating the grey knob at the bottom of the BSMS.
• Toggling on the FINE key causes the value to change more slowly as you turn the knob.
• A display in the middle of the BSMS shows the starting value on the left and the new value on the right.
• To accept the new value, press the STDBY key, or toggle on any other value setting key.  This automatically turns the light out on the key you were adjusting.
• Alternatively, if you toggle off the function you were just adjusting without accepting the adjustment, then the value will revert to the starting value.

To manually set the deuterium lock:
 
1. Make sure that the LOCK key on the BSMS is off and the SWEEP key is on.
2. On the xwinnmr menu bar click <Windows> and <Lockdisp> to open the lock display window.
• A trace should appear moving back and forth across the screen alternately tracing out a red and white decaying sine wave.
• If you see a flat line, the signal may just be off the screen.  Proceed to the BSMS FIELD adjustment below to bring the signal to the center of the screen.
• If the traces are both the same color, click  <mode>  in the Lockdisp window to get them to become red and white.
3. Toggle on the BSMS FIELD key and adjust the field to the center of the lock display window.
• Adjust until the largest white peak and the largest red peak are straddling the center grid line.
4. Toggle on the BSMS LOCK PHASE key and adjust the phase of the signal.
• The peaks should be adjusted to be of approximately equal height.
• A flat baseline is more important than peaks of equal height.
5. Adjust the intensity of the signal by changing input power and/or gain:
• Increasing power increases signal relative to noise, but too much power gives an unsatisfactory effect called saturation in which the sample can not dissipate the input energy through relaxation. In gross saturation, the signal will be unsteady.  In a mild state of saturation, if you slightly increase power the signal will initially increase but then drift back down.  The power must not be set high enough to observe these effects.
• Alternatively, increasing gain increases the signal strength without inducing saturation. But increasing gain has the undesirable effect of also increasing the noise.  So one wants the highest lock power short of saturation, and then the approproate lock gain setting to give the desired signal strength.
• First you will set them approximately, and then you will activate an automatic procedure to refine the adjustment.
1. Toggle on LOCK POWER, and adjust to a setting of -10 on the BSMS display.
2. Toggle on LOCK GAIN and adjust till peaks are near top of the lockdisp window (a value of about 120).
3. Accept by pressing STANDBY.
4. Press AUTOLOCK.  The autolock key will flash until the refinement is finished, and then the light on the LOCK key will come on.
• The instrument is now locked, meaning that it is automatically compensating for frequency drift.
• The lock display window will change to a signal moving back and forth in a  flat line at the position of the previous peak signal.  You will use this display for shimming below.
• If at any time the light on the LOCK key starts flashing, you have lost the lock.  You will then have to go back and reestablish it as above.  Most typically, this would happen if you lifted the sample out of the magnet for some reason.

Shimming the magnet

Shimming is the process of adjusting a series of small room temperature coils in the bore of  the magnet that function to cancel small disturbances in the homogeneity of the magnetic field induced by the sample. Failure to properly shim causes the peak shapes to deviate from the sharp lorenzian shape required to resolve closely spaced signals. Since the water signal is observed in the calibration steps below, one should achieve at least a crude shim at this point. If the sample has not come to thermal equilibrium yet, you can defer the fine adjustments until after the sample has come to thermal equilibrium and just before the experimental acquisition is about to begin. The shim adjustments for a particular configuration of sample and sample tube are saved in a shim file. Shimming is therefore started by retrieving a shim file for the closest available match to the current sample configuration (type of NMR tube, solvent, sample height). One should always start by reading in a shim file, otherwise the shimming process will be quite laborious.

Gradshim -- a new function was recently introduced that will use field gradients to measure the field at different points on the Z axis and automatically adjust the Z shims, including higher order shims, to make it homogeneous.  The Avance machines will either do shimming only on the Z axis (1D); or on all 3 axes (3D shim).  The following instructions are for 1D autoshim.  3D autoshim instructions will be added later.   With the 1D autoshim procedure, it is recommended to still read a shim file, then do the 1D gradshim.  For final shimming you will need to do the X and Y part of the manual shimming procedure, then repeat the gradshim program.  Note: do not confuse running the gradshim program with pressing the button on the BSMS named AUTOSHIM.  We do not use the latter function.
 

1. Type "rsh" (read shim) to read in a shim file.
• By default, xwinnmr keeps all shim files in the same system directory.
• The rsh command will produce a list of shim files to choose from.
• By convention, users add their initials as the extension when they write shim files to keep track of which are theirs.
• If you have previously saved a shim file for a similar sample, reload the file that you previously saved (by clicking on it).
• Otherwise, use aph_ub18shi (This is the shim file for an 18 mm thin-walled tube made by Shigimi).  This will eventually be replaced by a shim file (probably tw18mmshig.ref) with a ref  extension.)
• If you are using a different kind of tube, or a nonaqueous sample, or a different sample volume, you will have to ask someone to identify a more appropriate shim file.
• If someone gets a particularly well shimmed 17 or 16 mm sample,  we may get them to submit their shim file to the ref set.


The elementary shim operation is to depress a 2 keys  sequence on the BSMS that activates adjustment of a particular shim circuit.  The two keys will be one of the set (ON AXIS, X, Y), and one of the set (Z⁰, Z¹, Z², Z³).

If properly activated, the lights on the chosen pair will be on, and the others off.  The shim name will then be displayed on the BSMS display.  Once the selected shim circuit is activated,  turn the knob until the signal in the lock window is maximized.   Changing the shim too rapidly could conceivably cause loss of  the deuterium lock, in which case you should go back to the lock power and gain adjustment and reestablish the lock with AUTOLOCK.   If the signal goes off the top of the screen, just decrease the lock gain or lock power.  (Leave the instrument locked on during these adjustments; ie. don't touch the LOCK key).

The problem with shimming is that the different coils interact, such that optimizing one knocks the others off their optimum.  For a more extensive treatment of shimming see here.  For the lower order shims, one interatively optimizes them in the sequence specified below and expects to achieve a global optimum.  For higher order shims (eg. Z⁴ and above), the system can get trapped in local optimums.  For these one has to arbitrarily change the higher order shim, and then judge it by reoptimizing the lower order shim.  The only interaction that might benefit by this procedure among the shims adjusted below is the Z¹ Z³ pair.  The higher order shims should not need readjustment if:

• you have started with the correct shim file
• you have not pushed the plunger too far down into the tube
• you have properly centered the sample
• you have not included bubbles; Note: bubbles can form after you put the sample in the instrument, particularly if it came out of the refrigerator.


The 1D gradshim program automatically handles the interactions among the z shims, including higher order z shims.  You need to be aware that there are interactions between the X and Y shims and the Z shims.  Therefore after adjusting the X and Y shims, you should repeat gradshim.

If shimming is going badly, take the sample out and look for bubbles and such now.  If you have to remove the sample from the instrument at the final shimming step, then you will have to repeat essentially the entire procedure.

Step-by-step convergent shimming operation (replace steps 1-3 by running 1-D gradshim, or replace all of it by 1D followed by 3D gradshim):

1. Select the ON AXIS , Z¹ circuit and adjust the deuterium signal in the "Lock Window" to a maximum using the same round knob as before.
2. Maximize  ON AXIS, Z²_.
3. Maximize ON AXIS,  Z³.
4. Repeat the last 3 steps in order two more times.
5. Maximize X, Z⁰ .
6. Maximize Y, Z⁰_.
7. Maximize X, Z¹_; then X, Z²; then X, Z³.
8. Maximize  Y, Z¹; then Y, Z²; then, Y, Z³ .
9. Repeat ON AXIS; Z¹; ON AXIS, Z²; ON AXIS, Z³; X, Z⁰; Y, Z⁰.
10. You should not be seeing much improvement after the initial "On axis" optimizations.  If you are still getting improvements, then you should consider cycling through the shims again.
11. Type "wsh"  to save the shim file you have created and give it a name when prompted.  It will be especially useful to have the shim file saved if you later try to adjust higher order shims.
• If you owned the file to start with (meaning you originally wrote it) then you could use the same name as before and overwrite the previous file if you choose.  If you don't own the original file, you must give a new name.
• Eventually this directory will become overcrowded.  You can delete your obsolete shim files with the "delsh" command.


Running gradshim:

• Type gradshim to xwinnmr.
• The first time gradshim is run there is a set up process:
• Agree with the setup process and click "seen" to ignore all the error messages.
• For both 3D and 1D buttons (top of window) enter your username, but leave the disk as "u".
• Create an iteration control file (the following is for 1D shimming).
• <edit> <iteration control> <new>
• Give the file a name, ie sch5step18
• Add the following 5 steps:
• step 1, lowz, size=18
• step 2, lowz, size=18
• step 3, midz, size=18
• step 4, highz, size=18
• step 5, highz, size=18
• <save>
• The size is the length of the sample chamber.  The units don't exactly correspond to mm, but 18 is a good entry for an 18 mm chamber.  If these sizes are set too high, the procedure may fail to converge on a homogenously shimmed field.
• Thereafter, your iteration control file should be selected by default, but if not, select it from the list.  The * button expands the list of available iteration control files.
• <Start gradient shimming>
• It should say that it will do 5 iterations.  There is a bug in the AMX 500 version that causes it to sometimes only do 1 iteration.  If so, manually select <start gradient shimming> to make it iterate as necessary.
• A plot will be displayed of the frequency (as a measure of the field) over vertical coordinates with 0 at the center of the sample.  The size you have selected is in a different color.  The field is homogeneous when the central colored part of the plot is vertical.
• You may have to repeat gradshim to get the field homogenous.  First delete the previous plot window, else the new lines will be added to the old ones.  If the line has a hook on one end that can't be straightened, or appears to be unstable in its effort to converge on a verticle line, your sample chamber may not be centered properly.

Tuning the Probe

In order to operate with maximum sensitivity, the frequency (tuning) and the impedence (matching) of the RF coil should be adjusted for the nucleus under observation.  There is a separate RF detector for each kind of nucleus, and in this case the relevant one to tune is the proton coil.

Tuning requires setting a few parameters.  A dataset with the appropriate parameter files  has been created to set these for you.  If you have not already done so, you should first create a copy of the tuning dataset in your own directory under your own ownership.

To obtain the tuning dataset from another directory:
 

1. Under the xwinnmr <File> menu, open the <Search> window.  This will activate the Portfolio Editor dialogue box.
• A Portfolio is just the list of dataset you have chosen to look at in this session.
2. Select the /hinck disk drive (DU), and the user named "ref".
• This is a set of reference datasets now being set aside for new users to copy.  You may also copy a dataset from another user by this same procedure.
• If you do not see the /hinck disk drive, then set up the list of file locations.
3. Select tune1h.ref 1 1 either by double clicking on the file name (in the upper right portion) or by single clicking and then clicking <append>
4. Read in the file by double clicking on the name in the lower portion or by single clicking and then clicking <apply>.
5. Type "edc" to xwinnmr.
• This will activate a routine to copy the dataset and its associated parameters to your own directory (and prepare to receive data from an experiment #1).
• Once copied you will access the dataset from your own directory as indicated below.  This will protect you from confusion about finding the right dataset, and also in some circumstances from "write permission" errors caused by executing some command that wants to update the dataset but can not do so because you do not own it (have write permission).
6. In the edc dialogue box  change DU to the disk unit containing your user directory (/hinck or /nall), and fill in your own username.
• Leave the experiment number and process numbers as 1.
• Users often replace the extension with their own initials.  These instructions will refer to the files as tune1h.xxx hereafter.  To avoid subsequent confusion, one should reserve other changes to the name of the tuning dataset to mean that the tuning parameters within the dataset had been changed.
7. Click <save> in the edc menu to create the dataset in your own directory.


Or -- to access the parameter file in the tune1h.xxx dataset from your own directory:
 

1. As above, use xwinnmr <file><search> to find the dataset by specifying the disk unit (DU) that your directory resides on, your own username, and the dataset name that you had saved it under (presumably tune1h.xxx, with xxx as your initials).
2. Use <append> and <apply> as above to activate the dataset.
• Unlike the datasets with which you will actually collect data about your sample, there is no reason to change the experiment or process number attached to tune1h.xxx.  Doing so would only create redundant directories.
Carry out the tuning and matching
 
1. The <apply> operation above will have caused the tuning experiment to become active.
• Being "active" means that the parameters are loaded into memory and ready to be used by instructions that will be sent to the acquisition computer.
• The name of the currently active experiment is always displayed on the upper bar in the xwinnmr window.
2. With the cursor in the xwinnmr window, type "wobb" and enter. (you will be told that the display is in the acquisition window when wobb is complete).
3. Type "acqu" then enter (or go to the "acquire" menu and open "view fid") to open the acquisition window.
• This will cause a frequency sweep in the vicinity of the water proton peak.  A green V-shaped line will appear on the screen on a blue x- and y- axis representing the reflected energy vs. frequency.  If the blue axes do not appear, go to the unix shell from which xwinnmr was launched and type <control backslash> to reset the graphics display.
• To tune the probe, you will have to go to the magnet. Since you can not see the console from there, there is an LCD display to guide the operation on the preamp, which is located on the floor behind the magnet.
4. Locate two yellow capacitors (pots) on the underside of the magnet.  One is marked T and one is marked M.  Yellow encodes the hydrogen coil.
5. Using the tool connected by a chain, turn the pot marked T until the lit up LCDs in the horizontal row are centered.
6. The capacitor should turn easily.  Do not force it. If it doesn't turn, you have reached the endpoint.  Never force the capacitor past this point, else you will damage it and the probe will have to be sent back to Bruker for repair.
7. Then turn the pot marked M to reduce the height of the vertical row of lit LCDs.
8. The two pots interact, so go back and forth between them optimizing the tuning.
9. When properly tuned, there should be a single LCD lit in the center of the horizontal row and two or three lit at the bottom of the vertical row.
10. Return to the terminal and type "halt" to stop the acquisition ("halt" stops the acquisition after that pulse is finished.)
11. Click <return> on the acqu menu to exit acqu.

Calibrating the ¹H 90º pulses

The HSQC experiment will be conducted under the control of a pulse sequence program file named hsqc_fb.xxx.   The pulse program file itself is located in a system directory, and its name is stored as a parameter in the dataset also named hsqc_fb.xxx.  You will need to edit the pulse sequence program itself, so you must have ownership of a copy of that file as well as the hsqc_fb.xxx dataset. If you do not already own a copy of the hsqc_fb.xxx files, then you will need to make these copies first.

To obtain the hsqc_fb.xxx dataset/pulse program for the first time:
 

1. The first time you run a hsqc_fb experiment, you will obtain a copy of the dataset from the ref directory or another user with <file><search> as before.
2. Type "edc" to xwinnmr and save the active dataset under your disk unit and username, and with experiment number 1 and process number 1.  You may wish to change the extension to your initials, and alter the dataset name to include a designation of your protein.  The dataset will be further prepared to receive data from your first experiment below.
• You can type "pulprog" to xwinnmr to learn the name of the version name of the pulse program currently stored as a parameter in this dataset (probably hsqc_fb.ref).
3. To make a copy of this program under you own ownership:
1. Type "edcpul" to xwinnmr.
• This activates the vi editor and loads the pulse program indicated by the pulprog parameter in the active dataset.
• Remember that there is a command to tell xwinnmr which editor to use.
2. Type :w hsqc_.xxx <enter> to the edcpul window.
• The leading colon causes the rest of the command to appear on a command line at the bottom of the edcpul window.
• xxx means your initials.
• There must be a <space> between the w and the filename.
• This writes a copy by the name hsqc_fb.xxx which you will own.
3. Exit edcpul with by typing :q! in the edcpul window.
• If you were trying to copy a pulse program other than the one pointed to by pulprog, the manual suggested to me that you could use the alternative to edcpul:  edpul <filename>, then :w <new filename>.  I tried this, and edpul wouldn't allow any write operations.  Edpul is apparently only provided to look at other pulse program.  If you need to copy some other pulse program, transiently set pulprog to it, and then copy it with edcpul as above.
4. Then type "pulprog hsqc_fb.xxx" to xwinnmr where hsqc_fb.xxx is the new version of the pulse sequence program that you just made.  This will reset the pulprog parameter of the active dataset to the newly made copy of the pulse program.

To create an hsqc_fb dataset from one you have already used:

1. If you have already used the hsqc_fb procedure, you may prepare for a new data acquisition from a copy in your own directory.
2. Open your previous copy with <file><search><append><apply>.
3. Use "edc" and save  with the next higher experiment number.
• If you do not change the experiment number, then the data eventually gathered in this experiment will overwrite the data obtained in your previous experiment.
• The process number remains 1.  You will change this only in the processing step and only if you process the data more than once.
4. Verify with "pulprog" that it points to a copy of the pulse program that you own.  If not, fix it along the lines described above.

Adding a title and documentation to a dataset

To remind yourself about what data is in a dataset, you can give [each] experiment a title.

1. Type "setti" to xwinnmr.  This opens the vi editor to create a file describing the dataset however you like.
2. Exit the title editor with :wq
3. The command "title"  <yes> to xwinnmr will now display the first line of this title file below the name of the dataset in the window bar.
• ["title" <no>, however, didn't seem to remove it.]
• We are considering adding documentation to the ref datasets by this mechanism.  The documentation wouldn't be copied with the dataset, but we are exploring if the command "setti" would allow you to read the documentation before you copied the dataset.

Examine the hsqc_fb pulse program comments:

 
As a pulse is delivered, the resulting signal peaks and ebbs as a sine wave as a function of time and RF power. The time to reach the first maximum is called a 90 degree pulse.  A pulse of twice that time is called a 180 degree pulse and produces zero signal.  The estimated time and power required to produce a 90 degree pulse are recorded as comments in the pulse  program.  However, the exact duration required for the 90 degree pulse varies with the nature of the sample.  So one has to run a calibration procedure to determine the true 90 degree pulse time for your sample and manually enter it as a parameter to xwinnmr. Actually there are several different pulses that will require calibration, and the hsqc_fb pulse program contains instructions about what will need to be done.
 
 
1. You may examine or edit your copy of the pusle program (hsqc_fb.xxx)  referenced by the active dataset with the xwinnmr command "edcpul".
• Alternatively, "edpul <filename>" will allow examination of any pulse program that you know the name of (but not editing or copying).
• The comments at the top of the hsqc_fb file are as follows (only without the line numbers) [This listing is obsolete; it refers to the old AMX 500 pulse program.  References to hl1,2,etc will be pl 1,2, etc on the avance machines.  References to shaped pulse parameters starting with tp will be sp on the Avance system.  Nominal pulse times and attenuations will be different on each machine.  Refer to the comments in the pulse program on the actual machine you are using].
1. ;Over-Bodenhausen 1H-15N correlation with optional watergate,
2. ;water flipback, and 13C decoupling during 15N t1.
3. ;REF:  Grzesiek & Bax, JACS, 115, 12593 (1993)
4. ;
5. ;history
6. ;hsqcG.dif with Piotto Watergate H2O suppression scheme.
7. ;States with gradients:
8. ;Modified from andy's pulse program 10/25/93
9. ;installed on cber amx-500, 1/99
10. ;installed on uthscsa amx2-500, 12/99, phase cycle modified
11. ;
12. #include "AMX_APH.incl"
13. #define WG  ;watergate
14. ;#define CARBON  ;13c decoupling for 13c/15n samples (OPTIONAL)
15. ;#define FLIP_BACK ;flipback
16. #define TWO_D
17. ;#define ONE_D
18. ;
19. ;PROTON (channel f1 or "t") - Carrier on H2O (500.1336258)
20. ;p1 - hard 1H 90 @ hl1 for excitation, typically 8.5 us at hl1=2dB
21. ;p2 - soft 1H 90 @ hl2 for watergate, typically 1000 us at hl2=46dB
22. ;     MUST ADJUST PH14 & PH18 (See phase table below for details)
23. ;p3 - seduce1 shaped 1H 90 @ tp0 for OPTIONAL water flipback,
24. ;     typically 2000 us at tp0=47dB (set tpname0=seduce1.jc, tpoffs0=0)
25. ;     MUST ADJUST PH13 (See phase table below for details)
26. ;
27. ;CARBON (channel f2 or "d") - Carrier set according to f2list freq list (Ca,CO)
28. ;p5 - semiselective 13C 90 @ dl0, MUST be 28.5 us (typically at dl0=21 dB)
29. ;     (should give null at ~ 121ppm*125Hz/ppm off resonance)
30. ;
31. ;NITROGEN (channel f3 or "db") - Carrier set center of amide nitrogen (ca 118 ppm)
32. ;p7 - hard 90 @ dbl0 for excitation, typically 40 us at dbl0=11dB
33. ;p31- soft 90 @ dbl1 for cpd decoupling, typically 180 us at dbl1=23dB
34. ;     (set cpdprgb=waltz16.31)
Note: we are expecting to deposit a more heavily documented version of hsqc_fb.ref in the near future.
 
• Features of this file to note are as follows.
• Lines 13-17 define whether different techniques will be used or not during the experiment.  If there is a semicolon preceding the define statement, then the line is "commented out" and the technique will not be used.
• Line 13, watergate, should be uncommented
• Line 14, ¹³C/¹⁵N decoupling, should be commented out.
• This assumes that your sample was not labeled with ¹³C.  If it was, you will have to activate the decoupling technique to prevent the presence of the ¹³C from splitting peaks in your ¹⁵N spectrum.  In this case there will be a carbon pulses added to the program which will also need calibrated. Carbon pulses have to be calibrated on a ¹³C NaOAc in D₂O standard before placing your protein sample into the magnet. Nitrogen pulses are also used by the program, but need not be calibrated every time unless some problem with the machine has been encountered.
• Line 15, flip back, governs whether a water suppression routine named flip back will be used.
• Flip back suppression should be used under conditions of high amide exchange (pH > 6.0, or high temp.)
• Good quality spectra can be achieved with flip back off under conditions where amide exchange is not as rapid.
• A disadvantage of flip back is that its parameters must be more closely calibrated or the residual water peak will be large and distort the data.
• Since the reference copy obtained above has flip back suppression commented out, one will need to edit the file and save it to turn it back on. If you edit the file, be sure to use one of the write commands below to save your edits before you quit and close the window.
• edcpul has activated the vi editor.
• Bare bones vi editing commands are (case sensitive):
• arrow keys -- position the cursor
• x -- delete character at cursor
• i -- start inserting at cursor
• A -- start inserting at end of line
• <esc> exit insertion mode
• /<string> -- search for <string>
• :w -- (<colon>w) save the edited file
• :w <space><filename> -- save edited file with a new name
• :wq! -- write and quit
• :q! -- quit without saving the edits
• Line 16 collect 2D data should be uncommented.
• Line 17, collect 1D data should be commented out.
• Lines 21,22,24,25 give estimated values for 3 pulses that must be calibrated.
• Lines 23 and 26 warn you that there are some phase parameters that must be calibrated and then directly edited within the pulse program before it is run.  Those will be dealt with further below.  The comments about PH14 and PH18 are obsolete.  These edits have been replaced by setting the parameters phcor14 and phcor18.  The .ref copy of  hsqc_fb will soon have its comments updated.
2. Write down the estimated parameters for the proton pulses to be calibrated.
3. Exit edcpul (remember :q! exits without saving edits.  :wq! exits with saving.  If you own the file, you can't get in trouble with :wq unless you want to undo your edits.  On the other hand, if you are looking at a file that you do not own, :wq will cause a write permission error, even if you didn't change the file.  Write permission errors do not crash the editor, so you can easily recover from this error.)

Calibrating the ¹H 90º rectangular pulses (p1 and p2)

1. The 3 proton pulses are calibrated using a separate calibration dataset named calib1h.xxx.  The first time you use calib1h you will need to obtain a copy of the calib1h dataset, make a copy of the calib1h pulse program, and update the pulprog parameter to point to your version of the pulse program, just as you did for hsqc_fb.   Otherwise you can load calib1h with  <file> <search> <append> <apply> from your own directory.  Since you will not save data from this experiment, you can always leave the experiment number =1.
• We encourage you to obtain the .ref copies rather than some copy from another user.  Calib1h is used for many different kinds of calibration, and you could easily get a copy with an altered parameter that we are not mentioning in this document. The result could be strange and unanticipated behavior when you attempt your calibrations.
• Check sfo1 and sfo3 (by typing sfo1 and sfo3 as commands) against the current best frequency estimates for 1H (water), and 15N (mid amide) posted on the console.  Change them if they don't match.
2. Calib1h runs a pulse sequence to observe water protons and optimize the pulse time by trial and error.  With the calib1h dataset active, use edcpul to view your version of  the calib1h.xxx pulse program.  The first lines look as follows (additional comments may be added soon the the ref copy):
• ;calib1h.aph

;
#define NORMAL
;#define WATERGATE
;#define PRESAT
;#define SHAPE
• Calib1h can calibrate 4 different kinds of pulses.  p1 and p2 of hsqc_fb are normal (rectangular -- often called "square") pulses and so to calibrate them the #define NORMAL line should be uncommented and the other 3 define lines should be commented out as shown above.  If this is not the case, edit the file with vi commands and save it by :w (saves but does not exit).
• Note that you will repeatedly have to edit this program and then run it.  You can leave the edcpul window open, but you have to save each time  before you run because the program that runs is the one written on the disk, not the one in the editor's memory. [true?]
• If you get a permission error when you try to write, it's because you failed to make your own copy of the calib1h.xxx pulse program or failed to set your dataset pulprog parameter to your copy of the pulse program.  Go back and do those steps now.
3. Look into calib1h with edcpul to see what commands it uses to issue the normal pulse.
• The relevant lines are
#ifdef NORMAL
  10u hl1
  p1 ph1:r
#endif
• So you will have calib1h issue a pulse like the hsqc hard pulse (which  also uses p1 and hl1 parameters, but that correspondence doesn't always occur) by filling in p1 and hl1 for calib1h.
• These lines mean
• If NORMAL is defined at the top of the program, then do the following
• Wait 10 usec while the power is attenuated according to parameter hl1.
• Then issue a pulse of duration p1
• Set the phase of the pulse from the phase list ph1 (found at the bottom of the program) and correct it by phase correction parameter phcor1 (the :r specifies the use of a phase correction parameter with the same number as the pulse parameter).
4. Set the time and power parameters to the estimates for the hsqc_fb pulse #1 (known as the hard pulse).  The time is set by typing "p1" and entering 8.5u, and the power by typing "hl1" (lower case HL1) and entering 2. Note: these are direct xwinnmr commands for setting the parameters in the xwinnmr memory.  The pulse program will acquire them from xwinnmr when it runs.  Note: on avance machines, hl1 is replaced by pl1.
• If you leave out the units, microseconds is assumed for pulses. (Seconds is assumed for delay times, although you will not alter delay times in this setup procedure).
• The 2 dB is an attenuation value; high attenuation means less power.  This is called a "hard" pulse because it is at low attenuation, hence high power.  Pulses of this type generate the desired data; the soft pulses are for water suppression.
• As an alternative to typing specific parameter setting commands to xwinnmr, one can type the command "ased".  This displays most parameters used by the pulse program and allows any of them to be changed.   This is a generally useful command for monitoring the progress of your setup procedure.  However, phcor14 and phcor18 (mentioned below) are a couple of parameters that got left out of the "ased" display somehow, so you can't assume that "ased" will show you every relevant parameter.
5. Type "zg" (zero memory then go) to run the calibration program.
6. When the acquisition is finished, click <view as fid>, or type "acqu" to the xwinnmr window to open the acquisition window.
• A message will appear in the main xwinnmr window when the acquisition is finished.
• There is also a status line in the acquisition window.
• The acquisition window will display the fid (free induction decay).  This data is said to be in the "time domain" as opposed to the Fourier transformed data which is in the "frequency domain".
• The fid is the oscillating RF signal directly emitted by the stimulated nuclei.  It is represented as the difference between the observed nuclei and a reference signal (water hydrogens).  The signal decays over time.  Usually, there would be a collection of different frequencies superimposed due to the presence of hydrogen atoms with different chemical shifts.
• You are observing the hydrogens in water.  Since they are at the same frequency as the reference signal, they are said to be "on resonance".  This means that the oscillating sine wave is sampled at time intervals equal to its wave length.  In this case one can not see the oscillations, but rather just the gradual decay in intensity of the signal as a smooth line.
• The display appears as a vertical line, the top of which is the initial intensity observed; followed by a gradual decay.  Due to a phenomena called radiation damping, which will not concern us here, the decay of the water signal may reverse itself and drift upwards before decaying back to zero; the curve may also cross zero and make a peak of the opposite sign.
• You should see two such signals (one may be positive and the other negative).
• The one on the left is called the real signal or x signal; the one on the right the imaginary signal or y signal.
• If instead of thin lines, you see a colored in shape, then click <unsh>  in the lower left panel of the acquisition window.  (This means unshuffle.  Shuffle is a display that compares the real and imaginary signals).
• If you see a flat line, the scale of your display is off.  The upper left panel of the acquisition window has scaling buttons.  As you mouse over them, the message line at the bottom of the window will tell you what each one does.  Adjust the scale as convenient to see your display.
• These 2 signals represent the part of the RF sine wave arriving at two different detectors placed such that their signals are 90 degrees apart in phase.
• Your task will be to change the pulse time (p1) to maximize the sum of the absolute values of the initial intensities of these two signals.  To make this easier you will first adjust the phase of the stimulating pulse to shift all of the detected signal into one of the signals.  This is called shifting the signal into the real or the imaginary channels.  (It is convenient to focus on the the imaginary signal, since its initial deflection is in the middle of the screen. So I adjust phase to zero it; then add 90 degrees which makes it maximized. Whether the deflection is positive or negative doesn't matter.)
1. Adjust the phase of the stimulating pulse by typing phcor1 and entering a value in the range of 0.1 .. 90.0.
2. Type zg again and view the result.
3. Repeat until most of the intensity is in the imaginary signal on the right.
• The units of pchor1 are in degrees.  So, adding 90 would shift from maximum to zero, and adding 180 would shift from positive to negative.  So to zero out one of the signals (and maximize the other), you should make phcor1 some number less than 90.
• Getting most of the signal into the imaginary signal is good enough.
• The phcor1 correction only takes place for pulse programs that issue the pulse with a command of the form p1 ph1:r.  (see the lines displayed by edcpul above).  You will see an example of what happens if the programmer did not use this syntax when you calibrate pulse p3 below.
4. Write down the value of  phcor1.  You will use it again when calibrating the shaped pulse below.
 
7. Now you will adjust p1 (the pulse time) by trial and error.  Since it is easier to judge that the signal has been driven to zero than to judge if it has been maximized, you will start by entering ~4 times the estimated 90 degree pulse and by trial and error find a time that gives zero initial deflection. Note: 2 x the 90 degree pulse (180 degree pulse) and 4 x (360 degree pulse) should both give a zero signal.  The result of the 360 degree pulse is much less obscured by radiation damping.
1. Type p1 and enter 34u.
2. type zg and view the fid.
3. Repeat adjusting the time to make the initial deflection zero.
• Only the initial deflection (the nearly verticle line at the left edge of the imaginary signal) counts.  As the initial deflection approaches zero, it will become smaller, and the curve may then oscillate some.  If you go past zero, the initial deflection will reappear of the opposite sign.  It may be helpful to try several usec on either side of the anticipated zero point to see clearly what "not enough" and "too much" looks like.
• The optimal times for this and the other pulses increases with salt concentration in the sample in a coordinated way.  The p1 estimate in the hsqc_fb documentation above is about a usec high compared to the other pulse time estimates.  Long pulse times are undesirable, hence high salt is undesirable.  One can work up to about 250 mM, but lower salt is better.
4. Write down 1/4of the optimal 360 degree pulse time determined above.  This will eventually be entered into the hsqc_fb parameter list as the optimal 90 degree pulse time for the 2 dB pulse (p1 in hsqc_fb).  Note: the parameters you are working with in calib1h will be replaced when you load the hsqc_fb.xxx dataset.  So you have to store the optimized parameters on paper for a while.

 
8. Now calibrate hsqc_fb pulse 2 (also known as the soft pulse).
1. Type hl1 and enter 46dB.
2. Type p1 and enter 1000u
• Remember, Calib1h uses p1 and hl1 for all normal pulses, even though hsqc_fb has named this pulse p2.
3. Do the crude phase correction as above.
4. Find the time of the 360 degree pulse as above
• If the corrected 180 degree time differs by greater than 200 usec from the estimated value, then change the attenuation level by +or - 0.5 and see if you can get an optimal pulse time closer to the original estimate.  Ideally, one would do the adjustments by changing power rather than time, but the instrument can only change attenuation by 1/2 units of dB. Decreasing attenuation increases the RF pulse power and therefore decreases the required pulse time.
5. Write down 1/4 of the optimal 360 degree pulse to eventually enter as p2 of hsqc_fb.
6. Write down the attenuation level (if changed) to enter as hl2 of hsqc_fb.

Calibrating the ¹H shaped pulse (p3)

The 3rd pulse of hsqc_fb is a shaped pulse, meaning that the power will be raised and lowered according to a predefined cuve shape.  The statements describing it in hsqc_fb were:
 
 ;p3 - seduce1 shaped 1H 90 @ tp0 for OPTIONAL water flipback,
;     typically 2000 us at tp0=47dB (set tpname0=seduce1.jc, tpoffs0=0)

Note: The Avance instrument uses sp0 instead of tp0, and spnam0 rather than tpname0.


And the commands that issue the pulse in hsqc_fb are:

 
#ifdef FLIP_BACK
  10u tlo
  (p3 ph13):tp0
#else
  10u tlo
  d9
#endif
 
• The segment that issues pulse p3 only does so if you have defined doing the flip back procedure.  Else it executes a delay, which earlier in the program was set = p3.  So you do not have to do any of the shaped pulse calibrations unless you are doing flip back water suppression.  But without flip back water suppression, you still have to enter a reasonable value for p3 (2000 usec) prior to running hsqc_fb.  This dependency on setting p3 when not doing flip back may be removed in future releases of hsqc_fb.ref, in which case a comment at the top of the pulse program should say so.  It can't do any harm to set p3, however.
• The pulse is issued with phase list ph13, which will become relevant below.
• A power correction is applied to the pulse specified by the parameter tp0.  The comment warns you that this invokes a shaped pulse file such that the parameter tpname0 should be set to seduce1.jc, and another parameter tpoffs0 should be set to 0.  Presumably these later 2 parameters are already set in the initial parameter file for hsqc_fb (although you might want to check before running).  On the other hand, the calib1h parameter file may well not have them set.
1. You will have to edit the calib1h.xxx pulse program so that it will issue this shaped pulse.
1. Return to the edcpul window (still containing a copy of calib1h).
2. Put the semicolon back in front of the #define NORMAL line.
3. Remove the semicolon from the #define SHAPED line.
4. Write the file by :w
1. Check with "ased" that tpname0=seduce1.jc, tpoffs0=0, and tp0 = 47 dB.  If not, set them.
2. Note further down in calib1h.xxx the parameters it is using to define the shaped pulse:
#ifdef SHAPE
  d11 tlo
  (p3 ph3):tp0
#endif
1. Crude phase correction:
• The syntax for shaped pulses is a little different than we saw above.  During the preceeding delay time the machine is set into a low power mode by the command tlo, but the actual power isn't set until the parameter :tp0.  tp0 is the power (attenuation) parameter for p3.
• Note that instead of ph1:r, this procedure uses a phase specification of ph3 (without the :r) when delivering the pulse.  This is because the current version of the software can't support a power correction of the type :tp0 and a phase correction (:r) on the same command.
• This means that you can't use a phcor parameter to adjust the phase.  Instead you will do your crude phase adjustment by directly editing a line near the bottom of calib1h that says "ph3=".  This is called a "phase program" or "phase list".  A more extensive discussion is given below when the hsqc_fb phase list is edited..
2. The syntax is ph3=(360)x, where x is the phase change in degrees.
• The phase of this pulse is nearly always ~ -100 relative to the hard pulse, so a good first guess is the value of phcor1 you wrote down above when calibrating the hard pulse minus 100.
• The ph3= syntax doesn't tolerate negative numbers, so if you get a negative number, add 360.
3. Each time you edit  ph3= in calib1h you do a file write (:w) and then you do "zg" and view the fid to see if the signal is all shuffled into one of the channels.  As before, I prefer to zero out the imaginary signal (on the right), and then add or subtract 90 C to make that signal maximized. (keep the number between 0..360).
1. Once the signal is shuffled into the imaginary channel, find the optimum 90 degree pulse time.
1. Type "p3", and enter  4 x the estimated 90 degree pulse or 8000u.
2. Iteratively run zg and adjust p3 until you zero out the initial deflection.  As above adjust the attenuation (tp0) if that keeps the time closer to the initial estimate.
3. Write down 1/4 of the optimized 360 degree pulse as the optimized p3 time and write down the corresponding tp0 values for subsequent entry into hsqc_fb as the new p3 and tp0 parameters.

Determining the Phase Differences

Determining phase differences should no longer be required on the Avance class machines.  They are equipped with electronics to automatically compensate for phase variation at different attenuated power levels.  A procedure like that listed below might be used to verify the automatic phase correction if a malfunction were suspected.

The phases of the pulses that arrive at the sample must be coordinated.  Their relative phases are disrupted by the different levels of power amplification.  The following operation empirically determines a correction factor for each of the two soft pulses relative to the hard pulse.  These correction values will be required to directly edit into hsqc_fb before it is run.

To determine the reference phase.
 

1. Edit the calibration pulse program back to the "#define NORMAL" used for the square pulses.
2. Type "phcor1" then enter and set it to 0
3. Type "p1" and then enter and set it to 1/4 of the value determined for a 90º hard pulse.  Using 1/4 or 22 degree pulse will reduce the intensity of the signal and reduce distortions to the peak shape caused by truncation of signals that are too intense.
4. Type "hl1" and then enter and set it to the value determined for a 90º hard pulse.
5. Type "zg" then enter to acquire an fid.
6. Type "ef" (exponential multiply then fourier transform) then enter
7. Click  <Phase> on the left side of the window (if it’s not there, click <return> first at the bottom of the left side, and <return> on the menu that pops up).
8. Click and hold on <PH0>  while moving the mouse to phase the peak.
• In transformed data, an unphased peak appears to have both a positive and negative component.  When it is properly phased, it has a symmetrical shape with a flat baseline on both sides, and is positive.
• You can more accurately judge the symmetry by expanding the peak on the horizontal axis.  To do this, left click on the peak to attach the cursor to the line, and then middle click to the left and right of the peak.
9. Click <return> then <Save and return> on the menu that pops up. This stores the phase of the hard pulse as the reference for the next two measurements.


To determine the relative phase of hsqc_fb pulse p2.

1. Change the "p1" and "hl1" values to those determined for a 90º soft pulse. (hsqc_fb p2 and hl2).
2. Type "zg" and then enter to acquire an fid.
3. Type "efp" then enter (exponential multiply then fourier transform then phase with saved parameters)
4. Phase as before (by clicking and holding <PH0> and dragging).
5. Write down the PH0 value from the small box in the upper right hand corner (to 0.1 degree) as the phase difference for the soft square pulse relative to the hard square pulse. (usually about –90.00)
6. Click <return>, then <return> on the popup menu  — NOT <save and return> as before.
• If you mistakenly save p2 as the reference phase, then p3 will have its relative phase determined against p2 rather than p1 below.


To detemine relative phase of the shaped pulse (only needed if doing flip back):

1. Re-edit calibih as before to run the shape pulse.
2. Change the ph3= line in calibih to ph3=0, then do :w
3. Set "p3" and "tp0" to the values determined earlier.
4. Type "zg" then enter to acquire an fid.
5. Type "efp" then enter.
6. Click the phase button.
7. Click and hold <PH0> while moving the mouse to phase the peak.
8. Write down the PH0 value  (to 0.1 degree) from the small box in the upper right hand corner as the phase difference for the shape pulse.
• If the result is not around -90 to -100 degrees, then you probably replaced the hard pulse reference phase with the p2 phase by mistake.  Go back and set the reference phase again, then skip pulse p2 and determine the relative phase of the shaped pulse again.
9. Exit by <return> and <return> on the menu that pops up.


Checking the linewidth (by examining the transformed water peak).
 

1. Edit the calib1h back to the "#define NORMAL" used for the square pulses. :w
2. Type "hl1" then enter and set it to 2 dB.
3. Type "p1" then enter and set it to 2 (approximately a 20º pulse—a 90º pulse produces too much signal causing distortion).
4. Type "zg" then enter.
5. Type "lb" then enter and set it to 1. This applies a smoothing routine called "line broadening"
6. Type "ef" then enter.
7. Phase as before. Make sure that the peak is symmetrical. Asymmetry indicates the need for adjusting the Z⁴ shim.
8. Click <return> to leave the phasing mode, then <save and return>
9. Type "hwcal" then enter. This calculates the width at half-height.   A width of around 41 and a little asymmetry are acceptable.  Not all errors make the peak fatter.  A poorly tuned probe will make it thinner.

Final shimming

1. One should typically check the shimming again, particularly if the transformed peak appears asymmetrical or the peak width is unusually wide.  The 1D gradshim should effectively optimize higher order Z shims and eliminate most asymmetry. If gradshim was run earlier, and now it shows a big deviation from homogeneity, a bubble may have formed in your tube, and you will have to repeat most of the procedure.  Shimming may also change if the inital shimming was not done at thermal equilibrium.
2. Before attempting to improve a reasonably good shim, use wsh to save a shim file.
3. If you spin the sample (SPIN KEY on BSMS) then peak shape distortions due to X/Y shims will disappear.  This will confirm whether you need to work on higher order "On axis" shims or higher order X/Y shims.  Be sure to turn sample spinning back off.
4. Higher order shimming should be well handled by the gradshim program.
5. To make fine adjustments you must compare the detailed peak shapes.
1. Run the calib1h lineshape reference.  Phase carefully by expanding on the peak and zooming on the base of the peak (left click to attach cursor to line, and then middle click on either side of the peak).
2. <return> and <save> the reference phase.
3. Save for future comparison by "wpr 2".
4. Make a shim adjustment.
5. Run calib1h lineshape, and transform with "efp".
6. Type "dual" to superimpose the peaks.  Confirm that you want process number 2 of the calib1h dataset for comparison.  Enlarge and zoom to see the detailed effect on peak shape.
7. You can save additional trials as "wpr 3", "wpr 4", etc.
8. To recall a specified saved lineshape, use "edc2".
• "wpr n" saves processed data (the peak shape) to process number n of the calib1h dataset (a subdirectory named "n" under pdata).
• "edc2" gives an edc-like menu that allows you to define which process number you would like recalled to compare to the present peak shape when you issue the "dual" command.
• The recalled spectrum is #2.  The default color scheme makes the present spectrum (#1) green, and the recalled one purple.

1. Using <file><search><append><apply> Open the hsqc_fb dataset that you previously prepared to receive the data from this experiment. .
2. If you have not already done so, be sure to use edc to create a version with a new experiment number so that you do not overwrite data that you want to keep.
3. type "edcpul" to xwinnmr.
4. Make sure the 2D definition is uncommented, and 1D is commented out.
5. Check that flip back and carbon decoupling are set the way you want them..
6. You should ignore any comments about setting PH14 and PH18, however, check with the vi search command (/) that you have the pulse program version with the commands p2 ph14:r and p2 ph18:r.
7. Set phase program ph13 (only if doing flip back).
1. Find the line that says ph13=
• It is near the bottom of the program.  Use vi command G to go to the bottom.
2. Edit it to ph13=(3600) c, where c=1800 + 10 x the relative phase correction for the shaped pulse.
• For example, if the phased pulse was found to be -88.4 degrees relative to the hard pulse, then the line should be edited to ph13 = (3600)916

More generally for editing phase programs:

• If there is already a number there like ph13=2, these numbers are in 90 degree units.  Change it to ph13=(3600)1800, then make your subtraction.
• If there are several numbers there, subtract the same value from each one
• ie. ph13= 0 2 with a -90 degree correction would become:
ph13=(3600) 2700 900
• When there are multiple numbers in the list, the program will cycle through the list applying the pulse with a different phase on each scan.  This is done so that different scans will cancel out certain artifacts that would otherwise accumulate.  Therefore the total number of scans per fid (parameter NS, see below) should be an integral multiple of the length of each phase list appearing at the bottom of this program.
8. Save all changes and exit with :wq!.

• Spectral width; particularly in the ¹⁵N dimension.  The spectral width in the ¹⁵N dimension is typically set up to be 40 ppm, meaning that all spots in the 2D plot will have values within 20 ppm of the carrier frequency.  If there are signals outside of this range, they appear "folded" into the spectrum at a position + or - an integral number of spectral widths.  Folded peaks are often identified by being of inconsistent phase with the rest of the spectrum.  Folded peaks are confirmed by changing the spectral width, which shifts these peaks relative to the rest of the spectrum.  The spectral width in the proton dimension is set directly by parameter SW_h.  Spectral width in the ¹⁵N dimension is set indirectly by setting IN0, where SW = 1/(2*IN0).  Typically IN0 is 250 usec, giving a spectral width of 2000Hz or 40 ppm.
• d7 is a delay parameter that is intended to adjust the pulse sequence so that there is no first order phase correction in the proton dimension.  Currently, d7 is set as 100 usec, which is not quite right.  As a result there  is about a -45 degree first order phase correction required during the processing of the 2D spectrum.  (First order phase correction means that a phase adjustment is applied proportional to the resonance frequency of each peak).  Although this adjustment is easily made during processing, it is possible that d7 will be refined in the future.
• Nitrogen pulses:  p7 40 us, db10 = 11dB; p31 180 us at db11 = 23dB.  These pulse times and attenuation values might be adjusted based on an external calibration experiment.  The values don't change much, so people don't routinely recalibrate them.

1. Be sure hsqc_fb is active, else open hsqc_fb with <file><search><append><apply>.
2. Set parameters phcor14 and phcor18 according to the relative phase that you have determined for the soft pulse relative to the hard pulse (pulse 2 of hsqc_fb relative to pulse 1 of hsqc_fb).
• Negative numbers are OK for phcor parameters.
3. Type "ased" to xwinnmr. This displays and allows editing of most parameters used by the active pulse program.
4. Set the parameters p1, p2, hl2 according to the parameters you have written down.
5. Check that hl1 is 2 dB.
6. If you are doing flip back, set p3 and tp0 according to the parameters you have written down.  And check that tpname0 = seduce1.jc and tpoff0 = 0.
7. If you are not doing flipback, set p3 = 2000 usec.
8. Note the values of ns and l3 (lower case L3).
• ns is the number of scans per fid.  This generally controls the signal strength as well as the length of time that the experiment will take. NS=16 gives a good spectrum for an ideal sample in about 2 hours..  If the sample is known to be dilute, or the signal is expected to be weak because of suboptimal pH or temperature, then one would normally increase NS.  Typically one would either take the 2 hour acquisition, or go overnight (schedule permitting).  NS should be a multiple of 8 because of the phase cycling arrangement of the pulse program.
• l3 is the number of complex points in the ¹⁵N dimension.  This governs the resolution in the ¹⁵N dimension.  l3 should be 175, and you will not normally have reason to change it.  If changed, it will also alter the time taken by the experiment.
9. Check sfo1 and sfo3 against the current best frequency estimates for ¹H (water), and ¹⁵N (mid amide) posted on the console.  Change them if they don't match.
10. Accept the parameters and exit ased.
11. Type "expt" to xwinnmr.  This will tell you the total time that the acquisition will take.  If it's not what you expect, check the values of NS and l3 again.
12. At the command line type "1<space>TD<space> 350", (that's <one> <space>TD<space>350)
• This parameter (total data in 1st dimension) is used in the processing.  But if you do not set it to 2 * l3 now, it will cause a warning to be issued when you run the experiment.
13. Note the value of rg (receiver gain).
14. type "zg" and enter.

Checking how the experiment is going.

1. Wait till flashing lights on the RF console indicate that pulsing has begun, else you may crash the program by prematurely opening the acquisition window.
2. Type "acqu" then enter to open the acquisition window.
3. A small status window will be superimposed on the acquisition window. First it will count down through some negative number of scans.  These are dummy scans. The display is not meaningful until these complete.  Each 1 D spectrum is called an "experiment" in the information window.  When experiment 1 begins, the accumulated fid after each of the NS scans is averaged and will appear in the acquisition window at about 2 second intervals.  The axis around which the fid oscillates will itself undergo serpentine excursions around the horizontal center line due to the water signal.  The water suppression routines are designed to cancel these excursions out over a complete phase cycle of 8 scans.  If the individual fids move more than about 2 grid units away from the center line, then the signal is cut off (truncated) and water suppression will not work properly. Observe if the initial fids are displaced more than 2 grid units away from the horizontal center line.  If so, you need to improve you pulse times and phases.  If necessary, reduce RG to get the fid in range.  Another indicator of truncation artifacts is that after Fourier transformation, the baseline will oscillate to the right of the water position.
4. After the first 1D spectrum is complete, type "rser<space>1" (that's rser <one>).  This reads the fid for the first spectrum from a data file known as the serial file.
5. Type "ef" to tranform the data and then phase by clicking and dragging on <PH0>.
• One should have a complex pattern of positive peaks at the left with a flat baseline. This is a 1D 1H spectrum of  hydrogens attached to nitrogen in your protein (called a "nitrogen edited" spectrum).  There will be a residual water signal in the center that may have positive and negative components.
• If the water signal is large and distorts the protein spectrum, then there is something wrong with your water suppression.  Stop the acquisition ("stop") and go back and reexamine the way your pulses were calibrated and recorded, particularly phase parameters.
• Most commonly, the problem is with the flip back parameters.  You can confirm a flip back calibration error by restarting the program with flip back commented out and observing if the poor water suppression is relieved.
• Evaluate signal to noise.  The left-most part of the protein spectrum should have individual amide peaks visible above the noise on the baseline.
• If signal to noise is poor, you may consider stopping the experiment ("stop"), increasing NS (from a 2 hour experiment to a longer one), and starting it again.
• Alternatively, poor signal to noise ratio can come from poor shimming, or incorrectly calibrated pulse times.
6. If all seems well, allow the data to collect.
7. Click <return>, then on the popup menu click <Save as 2D and return>.  This saves phase information for use in the processing step.
8. The time remaining will be indicated on the acquisition window.

To allow the data to collect

1. You will not normally have the acquisition window active at this point, but if you do you must click <return> to return to the xwinnmr main display.
• The acquisition window is active if you see the protein fid dynamically changing every few seconds on the screen.
• Leaving this display on during a long acquisition crashes the program.
• Do not close the xwinnmr window itself.
2. Open a new unix shell. Go to the buttons on the desktop outside of the xwinnmr window and click <desktop><open unix shell>
3. Type "xlock"; this will put up a screen saver that can only be exited with your password.  This will protect your experiment from disruption by other users.
4. Come back when the experiment is over.
5. Use your password to exit the xlock
6. Confirm that xwinnmr indicates the experiment is over.
7. Use the LIFT button to retrieve your sample.
8. Toggle the LIFT button off.
9. Exit xwinnmr, and any other windows you have left around.
10. In the unix <desktop> menu, <logout>
11. Processing of the data will be described in another document. The computers in the NIS system can directly access the /u, /hinck, and /nall drives as "read only" devices by the names amx500_u, amx500_hinck, and amx500_hinck.

6/2/02 - some corrections contributed by Pete Walker were incorporated.
6/3/02 - SCH starting trying to flag some difference with the Avance machines, but the job is not done.